Restructuring of State-Owned Firms and Workers: Evidence from East Germany

Moritz Hennicke

Slides: https://AAoritz.gitlab.io/workers
Paper: http://hennicke.science/jmp.pdf

Daylight - Nightlight

Decarbonization and the Role of the State

• Subsidies and grants
• US Inflation Reduction Act
• State equity and ownership
• Nationalization of EDF in France 2022: "abundant production of carbon-free electricity" (Euractiv 2022)
• European Innovation Council R&D Funding
• China: SAIC Motor EVs

Can we learn from past state-led industrial transformations?

• Fall of Berlin Wall brought End of Fully State-Owned East German Economy
• Adoption of West German model $\rightarrow$ Probably most radical reorganization of industrial production and ownership in recent history

Timeline and Treatment Intensity

East German SOEs weighted by Initial Employment Over Time

Income, Unemployment and Life Satistfaction during the Transition

LP-DID: Difference-in-Differences with Local Projections (Dube et al, 2022)

• Advantages: Multiple continuous treatments, computation time, transparency
• Aggregate worker outcomes at year (t) - local labor market level (j) (defined by district and 2-digits industry) and weigh by initial market size
• Two staggered absorbing treatments: Log initial employment-weighted # of privatizations and liquidations
• Estimate for every event study coefficient $h$
\[ \underbrace{y_{j,t+h} - y_{j,t-1}}_{\text{Long difference}} = \underbrace{\beta^h_{LP-DID} \Delta Privatizations_{jt}}_{\text{First-differenced treatment}} + \underbrace{\tau^h_t}_{\text{Time FE}} + \epsilon^h_{jt} \]
• Restricting observations to either
• newly treated: $\Delta Privatizations_{jt} > 0$ but $\Delta Liquidations_{jt} = 0$
• or "clean control": $Privatizations_{j,t+h} = 0$ & $Liquidations_{j,t+h} = 0$
• Analogously for liquidations
• Identification assumptions: Parallel trends, no anticipation
• Limitation: General equilibrium spillovers

• [Robustness]

Robustness

• [Balanced] panel on local labor markets
• Selected treatment group: Only [stayers] in local labor markets
• Selected treatment group: Only [movers] across districts and industries
• [Matching] on pre-treatment observables: 1d industry, binary gender, age, college

DiD: Privatization Impact on Firms

• Exits: 32% of privatized firms do not survive first 5 years (Mergele, Hennicke and Lubczyk 2022)
• Data: Balanced panel from 1991-1997
• Estimate for every event study coefficient $h$ with annual firm ($j$) panel
\[ \underbrace{y_{j,t+h} - y_{j,t-1}}_{\text{Long difference}} = \underbrace{\beta^h_{LP-DID} \Delta Privatization_{jt}}_{\text{1st-diff. treatment indicator}} + \underbrace{\gamma^h \Delta y_{j,t-1}}_{\text{1st-diff. outcome lag}} + \underbrace{\tau^h_t}_{\text{Time FE}} + \epsilon^h_{jt} \]
• Control group: Not-yet treated and firms remaining state-owned (abstract from changes in federal, state or city ownership)
• Matched on state, 1 digit industry, privatization agency division
• Robustness: Public utility industries, Firms in municipalities that filed recommunalization requests
• Identification assumptions: Parallel trends
• Limitations: General equilibrium spillovers

Conceptual Framework

• Bargaining: Workers with private or state-owned enterprise following Arnold (2019)
• Introduce shocks to profits as argued by Demougin & Sinn (1994)
• Assuming normally distributed shocks to labor input and CARA
\[ \max_{L_j, K_j, w_j} \underbrace{(L_j w_j + (\bar{L}_j - L_j)v - \bar{L}_j v)^{\gamma_j}}_{workers} \\ \underbrace{(\lambda_j (F_j(K_j, L_j) + \tau_j L_j - r K_j - w_j L_j - \lambda_j \frac{\sigma^2}{2} L_j ))^{1-\gamma_j}}_{firm} \]
• Privatization might lead to downsizing due to
• Higher per worker subsidies $\tau_j$ to state-owned enterprises
• Superior risk-sharing possibilities of state $\lambda_j$

Marginal Value of Public Funds

• Hendren and Sprung-Keyser (2020)'s MVPF for a 1% reduction of (employment-weighted) privatizations (from average level)
\[ MVPF = \frac{WTP}{\text{Net Costs to Government}} \]
• Willingness to pay by workers
1. Avoided monthly household income loss
\[ -\beta_{LP-DID}^6 * \bar{inc} * 12 * 6 \]
• Net cost to the government
1. Decrease in annual public transfers (UB, social assistance etc)
\[-\beta_{LP-DID}^6 * \bar{transf} * 6 \]
2. Direct subsidies to SOEs:
   \[ -\frac{\text{All subsidies to SOEs 1990-1994}} {\text{Sum of annual employment in SOEs 1990-1994}} * 6 * \Delta (-\bar{ewP}) \]
3. No privatization revenue, tax revenue etc

Conclusion

• Can East Germany's transition inform current efforts to decarbonize industry?
• Reunification likely led to dip in GDP (Abadie, Diamond, Hainmueller 2015)
• Germany's East still lags behind its West
• Close to 1/2 of East German workers would have benefited from slower phase-out of state-ownership
• At what costs?

Source: Abadie, Diamond, Hainmueller (2015)

Employment and Dimensions of Life Satisfaction in East Germany

Source: Based on retrospective questions in SOEP (2017)

• [Robustness]

• [Robustness]

• [Robustness]